Aplastic anemia (AA) and other types of bone marrow failure have clinical and laboratory features consistent with an autoimmune pathophysiology, possibly incited by a virus. A majority of patients respond with hematologic improvement after immunosuppressive therapies. Our laboratory studies have focused on aspects of the immune pathophysiology of hematopoietic suppression, the nature of the viral antigen, and the mechanism of late clonal evolution of aplastic anemia to other diseases, including myelosyplasia and paroxysmal nocturnal hemoglobinuria (PNH). In addition, we test new therapies for marrow failure in research clinical protocols. One form of AA occurs after hepatitis, and we have completed analysis of combined clinical and laboratory data on 10 patients seen at NIH. Most were young males who suffered seronegative hepatitis followed by severe marrow failure. The hepatitis was characterized as non A non B non C and also non G; a role for the novel GBV- C/hepatitis G agent was excluded here and also in AA in general. There were strong HLA correlations within the hepatitis/AA population, especially to Cw7, B7, and DR7, DQ2, and Drw53, and all showed evidence of cytotoxic lymphocyte activation. Seven patients showed excellent responses to immunosuppressive therapy. Hepatitis/AA is an immunologically mediated marrow failure syndrome incited by an unknown virus. In studies of hematopoiesis in AA, we have measured stem cell numbers in a surrogate assay called the long-term culture initiating cell test (LTC-IC). All patients show a marked deficit in stem cell number on presentation that does not allow prediction of response to immunosuppressive therapy. LTC-IC also are qualitatively abnormal with a low capacity to form secondary colonies in culture. Although patients can recover blood counts without increasing numbers of LTC-IC, several years after treatment, some patients do have normal numbers of LTC-IC, suggesting repopulation within the stem cell compartment. Finally, we have established a knock-out model of hematopoiesis in PNH. Destruction of the PIG-A gene, required for expression of glycophosphoinositol (GPI)-linked proteins on the cell surface, adversely affected embyrogenesis in vitro; however, embyroid body formation was restored, and hematopoiesis from genetically defective cells was normal, after coculture of PIG-A- and normal cells.Intercellular protein transfer could be demonstrated. Studies with red cells also suggest that GPI-linked proteins can be exchanged by high density lipoproteins. For human patient samples, there were no differences in hematopoietic colony formation between normal and GPI-deficient cells. These results suggest that PNH hematopoiesis has a selective advantage in AA.